[PATCH 4/4] net/ark: improve Rx queue recovery after mbuf exhaustion

[Stephen Hemminger]

On Wed,  3 Sep 2025 17:28:45 -0400
Ed Czeck <ed.czeck at atomicrules.com> wrote:

> @@ -185,12 +184,12 @@ eth_ark_dev_rx_queue_setup(struct rte_eth_dev *dev,
>  	queue->reserve_q =
>  		rte_zmalloc_socket("Ark_rx_queue mbuf",
>  				   nb_desc * sizeof(struct rte_mbuf *),
> -				   512,
> +				   4096,
>  				   socket_id);
>  	queue->paddress_q =
>  		rte_zmalloc_socket("Ark_rx_queue paddr",
>  				   nb_desc * sizeof(rte_iova_t),
> -				   512,
> +				   4096,
>  				   socket_id);
>  

Not all ARM systems use page size of 4096. Some are larger.
The parameter changed is the alignment value, not the size.
The allocation is taking place on huge pages where the page size
is typically 2 Megabytes or larger. Why is this doing anything useful?